Design, fabrication, and characterization of a micro fuel processor

(1)

Design, Fabrication, and Characterization of a Micro Fuel Processor

by

Brandon S. Blackwell

B.S. Chemical Engineering, University of Notre Dame, 2002

M.S. Chemical Engineering Practice, Massachusetts Institute of Technology, 2004

Submitted to the Department of Chemical Engineering in partial fulfillment of the requirements for the degree of

Doctor of Philosophy at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY February 2008

Signature of Author ... .. ... . .... ...

Department of Chemical Engineering September 5, 2007

/7,/

Certified by ... .-. . ..:.. . .. ... .. . ...

Jensen Klavs F. Jensen

Warren K. Lewis Professor of Chemical Engineering Professor of Materials Science and Engineering Thesis Supervisor

Abcepted by.. ... ... ...

William M. Deen Carbon P. Dubbs Professor of Chemical Engineering

c

HVr

sýf

T Chairman, Committee for Graduate Students

SOFTEOH -OLOGy

MAR

0 6

2008

LIBRARIES

Design, Fabrication, and Characterization of a Micro Fuel Processor by

Brandon S. Blackwell

Submitted to the Department of Chemical Engineering on September 5, 2007 in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Chemical Engineering Abstract

The development of portable-power systems employing hydrogen-driven solid oxide fuel cells continues to garner significant interest among applied science researchers. The technology can be applied in fields ranging from the automobile to personal electronics industries. In order for fuel cell systems to outperform batteries, a method of chemically converting high-energy-density combustible fuels to hydrogen while maintaining high thermal efficiency must be developed. This thesis focuses on developing microreaction technology that minimizes thermal losses during the conversion of fuels - such as light end hydrocarbons, their alcohols, and ammonia - to hydrogen. Critical issues in realizing high-efficiency devices capable of operating at high temperatures have been addressed: specifically, thermal management, the integration of materials with different thermophysical properties, and the development of improved packaging and fabrication techniques.

A new fabrication scheme for a thermally insulated, high temperature, suspended-tube microreactor has been developed. The new design improves upon a monolithic design proposed by Leonel Arana. In the new modular design, a high-temperature reaction zone is connected to a low-temperature package via the brazing of pre-fabricated, thin-walled glass tubes. The design also replaces traditional deep reactive ion etching (DRIE) with wet potassium hydroxide (KOH) etching, an economical and time-saving alternative. A glass brazing method that effectively accommodates the difference in thermal expansion between the silicon reactor and the glass tubes has been developed. The material used in this procedure is stable at temperatures up to 710 oC.

Autothermal combustion of hydrogen, propane, and butane in excess oxygen has been demonstrated in ambient atmosphere and under vacuum. Hot spot temperatures of up to 900 OC

have been measured during autothermal combustion of propane in ambient and vacuum conditions. Experimental temperature measurements have been compared to steady-state temperature estimates, and show good agreement. Finally, a computational fluid dynamics (CFD) model has been developed to study the heat transfer properties of the microreactor. Using simplified reaction schemes from the literature, the model has successfully reproduced the results observed in the laboratory.

Thesis Supervisor: Klavs F. Jensen

(4)

(5)

Acknowledgments

Many people have made my time at MIT an enjoyable and educational experience. I would like to thank my advisor, Klavs Jensen, for his consistent support and direction. His passion for collaboration with outside parties has allowed me to develop relationships that never would have been possible otherwise. In addition, I would like to thank the members of my thesis committee, Professors Martin Schmidt, William Green, and Jeff Tester. Each provided unique insight and helpful recommendations as the project progressed. I would also like to thank Multidisciplinary University Research Initiative (MURI) and the Lincoln Labs Advanced Technology Concepts program for funding.

I have also had the opportunity to work alongside many talented students and post-doctoral associates during my time here. I would like to thank Smitha Matthews and Ben Wilhite for guiding me during my first few months in the fabrication facility. I would also like to thank Sravanti Kusuma and Gian Caviezel for their assistance in assembling the testing apparatus used for experimentation. Maurizio Rondanini played a critical role in developing and refining the solution grid employed by the CFD-ACE model described in Chapter 7. I also thank Kishori Deshpande for taking the time to discuss issues with my project, helping with the maintenance of common lab equipment, and her impromptu lab clean ups.

I also owe thanks to the entire KFJ research group. I could not have asked for a more welcoming environment in which to work. Specific thanks go to Jake Albrecht, Jane Rempel, and Jamil El-Ali for the water cooler breaks in 66-501. I would also like to thank Joan Chisholm and Alina Haverty for their patience in dealing with my complicated supply requests and for being great resources in general.

Most importantly, I would like to thank my family, to whom this thesis is dedicated. I thank my parents, Steve and Cheryl, for their love and support; my brother, Sean, for his weekend visits that forced me to get out of the lab; and I think my sister, Emily. Finally, I thank my wife, Kenzie. For over three years she tolerated the roller coaster ride - my frustrations and complaints, successes and celebrations - and provided unwavering support.

(6)

(7)

List of Figures

Figure 1-1. Sales growth of various rechargeable cell technologies. [5] ... 18 Figure 1-2. Battery schem atic. ... 20

Figure 1-3. Modified Ragone plot of energy density vs. specific energy for current technologies in (a) primary batteries and (b) secondary batteries. [5, 9] ... 21

Figure 1-4. Plot of electrical energy versus system weight for generators and batteries. Note

that the steps in the battery curve represent the incremental increase in system performance with an increase in number of discrete cells. [ 11] ... ... 23 Figure 1-5. Schematic of hydrogen-powered fuel cell operation. From [23] ... 27 Figure 2-1. Schematic of heat-loss pathways associated with a high-temperature fuel processor.

Process heat is lost via conduction, natural and forced convection from the exterior of the reaction zone, radiation from the reaction zone, and process heat lost due to forced

convection of the process gasses through the reactor. ... 41 Figure 2-2. Theoretical conductive heat loss from an 8000_{C fuel processor to a 25}0_{C heat sink}

along one tube for various materials. Commonly available tube materials and sizings were chosen. Thermal conductivities are assumed to be constant over the temperature range.... 45

Figure 2-3. Theoretical conductive heat loss from an 8000C fuel processor to a 250C heat sink

along one ultra-thin-walled capillary tube. Tube materials were chosen from those compatible with silicon microfabrication. Thermal conductivities are assumed to be

constant over the temperature range. ... ... ... 46

Figure 2-4. Exchange of radiation between two surfaces with differing view factors. As the hot and cold surfaces approach each other, F approaches 1. ... 49

Figure 2-5. Apparent emissivity, EA, of a hot surface in close proximity to a cold surface (view factor = 1)... ... ... 50

Figure 3-1. Schematic of Arana's suspended tube reactor (SgLRE I). The four suspended tubes

are each 200im wide by 4801am high. The overall die dimensions are 8 x 10 mm. [38]... 54 Figure 3-2. Photo and cross-section schematic of SpRE III, as designed by Nielsen [52]... 58

Figure 3-3. Schematic of the new suspended tube micro fuel processor (SpRE IV). ... 58 Figure 3-4. Schematic of multiple reactors operating in parallel. As each reactor is added, the

amount of channel volume and exposed catalyst increases by whole multiples of a single unit. The amount of external surface area (highlighted in red) is increased by a fraction of a single unit ... ... ... 60

Figure 3-5. Schematic of parallel opereation of three SpRE IV reactors to achieve autothermal endotherm ic reform ing... 61

Figure 3-6. Schematic of SpRE IV reactor used for TPV power generation... 62 Figure 3-7. Maximum tolerable pressure gradient across tube walls for various tube radii and

thicknesses. Calculations based on glass fracture strength of 117 MPa ... 68

Figure 4-1. Schematic of the fabrication process for the high-temperature reaction zone... 71

Figure 4-2. Schematic of the fabrication process for the reactor frame... 71 Figure 4-3. SEM micrograph of an etched structure showing convex corner undercutting [55].

... 73

Figure 4-4. Example comer compensation structure [56]... ... 74

Figure 4-5. Updated corner compensation structure [58]. ... ... 75 Figure 4-6. Photos of reactor microchannels after catalyst loading and subsequent cleaving.... 77

(12)

-Figure 4-7. Reactor setup for glass tube brazing procedure. ... 79 Figure 4-8. KOH etching results for original channel layout. ... 80 Figure 4-9. KOH etching results for modified channel layout. ... 81 Figure 4-10. KOH etching results for reactors employing newest comer compensation features.

... ... ... 82

Figure 4-11. Photos of completed SpiRE IV with silicon frame. Note that in (b) the sealant at the frame has been replaced with epoxy ... .... ... 82 Figure 4-12. Photos of completed SpRE IV without silicon frame ... 83 Figure 4-13. Two SpRE IV reactors bonded to form a stack... ... 83 Figure 4-14. Packaging of reactor for chemical testing. This setup allowed for testing in both

ambient atmosphere and vacuum at pressures as low as 16 mTorr ... 85 Figure 4-15. Assembled reactor for chemical testing. ... ... 85 Figure 4-16. Infrared thermometer experimental setup. ... ... 87 Figure 5-1. Calculated estimate of conductive heat losses from two tubes in SpRE IV under

isothermal, steady-state conductions. The 700-ipm capillary tubing has an inner diameter of 500 microns, while the 550-pm capillary tubing has an inner diameter of 400 microns... 90 Figure 5-2. Calculated estimate of convective heat losses from the silicon reaction zone in

SpRE IV under isothermal, steady-state conductions. Note that this estimate does not include convective losses from the sealant or capillary tubes. ... 92 Figure 5-3. Mean free path between collisions as a function of absolute pressure. Note that

760 Torr = 1 atm ... 93 Figure 5-4. Spectral emissivity of pure silicon. From [61] ... 94 Figure 5-5. Lower and upper limits of calculated radiative heat losses from the silicon reaction

zone in SpiRE IV under isothermal, steady-state conductions. Note that this estimate does not include radiation losses from the sealant or capillary tubes. ... 95 Figure 5-6. Comparison of heat loss pathways from the silicon reaction zone in SpRE IV under

isothermal, steady-state conductions. Note that this estimate does not include radiation losses from the sealant or capillary tubes. ... ... 96 Figure 5-7. Calculated estimate of convective and radiative losses from the glass sealant in

SpjRE IV under isothermal, steady-state conductions... ... 97 Figure 5-8. Percentage of generated heat lost via heated exhaust gases from SjpRE IV under

isothermal, steady-state conductions ... 99 Figure 5-9. Predicted steady-state heat loss and energy generation in SjpRE IV at 4000C as a

function of hydrogen inlet flow rate. The heat generation line corresponds to the full

combustion of fuel at a 2:1, oxygen to fuel stoichiometric ratio. ... 101 Figure 5-10. Predicted steady-state heat loss and energy generation in SjpRE IV at 4000C as a

function of propane inlet flow rate. The heat generation line corresponds to the full

combustion of fuel at a 2:1, oxygen to fuel stoichiometric ratio ... 101 Figure 5-11. Predicted steady-state heat loss and energy generation in SjpRE IV at 550'C as a

combustion of fuel at a 2:1, oxygen to fuel stoichiometric ratio. ... 102 Figure 5-12. Predicted steady-state heat loss and energy generation in SjpRE IV at 5500C as a

function of propane inlet flow rate. The heat generation line corresponds to the full

(13)

Figure 5-13. Predicted steady-state heat loss and energy generation in SpLRE IV at 7000C as a

combustion of fuel at a 2:1, oxygen to fuel stoichiometric ratio ... 103

Figure 5-14. Predicted steady-state heat loss and energy generation in SoRE IV at 7000_{C as a} function of propane inlet flow rate. The heat generation line corresponds to the full

combustion of fuel at a 2:1, oxygen to fuel stoichiometric ratio ... 104

Figure 5-15. Predicted autothermal steady-state temperature of SgLRE IV during complete

conversion of hydrogen... 105

Figure 5-16. Predicted autothermal steady-state temperature of SpRE IV during complete

conversion of propane ... 105

Figure 6-1. Experimental setup... 107 Figure 6-2. Autothermal, steady-state combustion of 40 sccm hydrogen with 2.5 mg of (a)

lwt% Pt on A1203 and (b) 5wt% Pt on A1203... . . . ... . . . 111

Figure 6-3. Temperature profile of SjRE IV as a function of hydrogen flow rate in ambient

environment. In all cases, a 2:1 stoichiometric ratio of oxygen to hydrogen was fed to the reactor, and 100% conversion was observed. Temperature measurements were made using an IR thermometer calibrated using Omega indicating lacquers ... 113

Figure 6-4. Temperature profile of StLRE IV as a function of hydrogen flow rate at 16 mTorr

vacuum. In all cases, a 2:1 stoichiometric ratio of oxygen to hydrogen was fed to the reactor, and 100% conversion was observed. Temperature measurements were made using an IR thermometer calibrated using Omega indicating lacquers ... 113

Figure 6-5. Autothermal, steady-state combustion of 13 seem propane with 2.5 mg of (a) lwt%

Pt on A1203 and (b) 5wt% Pt on A1203 ... ... . . ... 115

Figure 6-6. Temperature profile of SjtRE IV as a function of propane flow rate in ambient environment. In all cases, a 1.5:1 stoichiometric ratio of oxygen to propane was fed to the reactor and 91%+ conversion was observed. Temperature measurements were made using an IR thermometer calibrated using Omega indicating lacquers ... 116

Figure 6-7. Temperature profile of SjtRE IV as a function of propane flow rate at 16 mTorr

vacuum. In all cases, a 1.5:1 stoichiometric ratio of oxygen to propane was fed to the reactor. 97%+ conversion was observed for propane flow rates of 6 seem while greater than

99% conversions were observed for all other flow rates. Temperature measurements were

made using an IR thermometer calibrated using Omega indicating lacquers... 117

Figure 6-8. Autothermal, steady-state combustion of 9.5 seem butane over 2.5 mg of lwt% Pt

on A l20 3 ............. ... . . . . .. 118

Figure 6-9. SgLRE IV reactor after failure via homogeneous combustion of hydrogen. Note that the inlet tube has expanded significantly due to the high temperatures of the reaction flame.

... ... ... ... ... ... ... .... 120

Figure 6-10. SpRE IV reactor after failure via delamination of the glass sealant from the

reaction zone. Note that the failure resulted in a clean break along the interface where the braze met the silicon reactor. ... 121

Figure 6-11. SpRE IV reactor after failure via burning of the epoxy joining the outlet capillary

to the aluminum testing fixture. ... 121

Figure 7-1. Reactor geometry ... 123 Figure 7-2. Volume element mesh used to simulate catalytic hydrogen combustion in SpRE IV.

Note that the concentration of the mesh in the reactor channels, in the glass sealant, at the ignition points, and at the inlet and outlet ... 128

(14)

-13-Figure 7-3. Velocity profile in SgIRE IV for the combustion of 26.5 sccm hydrogen in 26.5 sccm oxygen . ... 134

Figure 7-4. Simulated hydrogen concentration as a function of position in SgtRE IV. For this

simulation, the hydrogen flow rate is set at 53 seecm. A 2:1 stoichiometric ratio of oxygen to hydrogen w as supplied to the inlet... 135

Figure 7-5. Comparison of (a) experimental temperature profile to (b) simulated temperature

profile for the combustion of 53 seem of hydrogen in 16 mTorr vacuum. A 2:1

stoichiometric ratio of oxygen to hydrogen was supplied to the inlet ... 136

Figure 7-6. Comparison of (a) experimental temperature profile to (b) simulated temperature

profile for the combustion of 16 seem of hydrogen in 16 mTorr vacuum. A 2:1

stoichiometric ratio of oxygen to hydrogen was supplied to the inlet ... 136

Figure 7-7. Steady-state propane concentration profile for initial propane flow rates of

(a) 6 seem, (b) 8 seem, and (c) 10 seecm. As the flow rate is increased, more propane reacts closer to the inlet. A 1.5:1 stoichiometric ratio of oxygen to propane was supplied to the inlet . ... 137

Figure 7-8. Comparison of (a) experimental temperature profile to (b) simulated temperature

profile for the combustion of 6 seem of propane in 16 mTorr vacuum. A 1.5:1

stoichiometric ratio of oxygen to propane was supplied to the inlet. ... 138

Figure 7-9. Comparison of (a) experimental temperature profile to (b) simulated temperature

profile for the combustion of 10 seem of propane in 16 mTorr vacuum. A 1.5:1

(15)

List of Tables

Table 1-1. Energy Properties of Common Fuels [10] ... ... 23

Table 4-1. Properties of Materials Considered for Tubing [59, 60] ... 78

Table 5-1. Orientation dependence of b and m in Equation 5-1 ... 91

(16)

Chapter 1 Introduction

Portable electronic devices - from cellular telephones to laptop computers to iPods

-continue to require greater amounts of energy for longer periods of time. While energy demands have increased, portable power technology has failed to maintain pace. The battery has served as the primary vehicle for powering portable electronics, and improvement in battery technology remains an active area of research. Alternative research efforts have focused on developing portable electric generators such as fuel cell systems, microengines, and thermophotovoltaic systems. Electric generators have the potential to reach higher levels of performance due to the fact that the combustible fuels used to power them have energy densities an order of magnitude greater than those of their battery counterparts.

Of the types of electrical generators mentioned above, fuel cell systems are advantageous due to their high energy density, ease of miniaturization, quiet operation, and lack of moving parts. Of the many available types of fuel cells systems, only a few are suitable for portable power applications due to size restrictions and the desire to avoid the use of hazardous chemicals. Polymer electrolyte membrane (PEM) fuel cells, direct methanol fuel cells (DMFCs), and solid oxide fuel cells (SOFCs) have garnered the most attention for portable power applications.

Hydrogen is the best-performing option among combustible fuels, but its use requires an effective storage scheme. A number of strategies have been investigated including cryogenic storage, compression in high pressure vessels, adsorption to carbon nanostructures, and chemical binding in reversible and non-reversible metal hydrides. Of these technologies, only borohydride-based systems have been proven to outperform batteries [1]. As an alternative to hydrogen storage, direct processing of liquid fuels has been investigated. The direct methanol

(17)

fuel cell, for example, generates power by directly converting easily stored liquid methanol. While direct methanol systems have been developed for practical use, their efficiencies lag behind those theoretically achievable in high-temperature solid oxide fuel cells. This thesis investigates a third option: on-board reforming of combustible fuels to hydrogen-rich streams via a portable fuel processor.

1.1

Motivation

1.1.1

Consumer Electronics

Since the inception of the integrated circuit - and especially over the last two decades

-the number and diversity of consumer electronics devices on -the market has swelled at a fast rate. The sales of laptops and wireless telephones have significantly increased annually since their introduction to the marketplace. Annual worldwide laptop computer sales doubled from 2002 to 2006, growing from 36 million to 72.5 million units sold [2, 3] . Wireless phone sales have also increased substantially, from 600 million units sold in 2002 to more than 1 billion units in 2006 [2, 4]. Portable power systems have the potential to be used in myriad other devices including music players, portable gaming devices, digital cameras, and camcorders, among others.

Growth has been observed not only in the market for portable devices, but in the energy demand of these devices as well. For example, laptop computers with brighter screens require more power to operate. The addition of audio file playback and internet browsing features in wireless phones has taxed current power delivery devices. Along with delivering more power, it is also desirable to deliver the power for longer lengths of time. A laptop computer that can operate continually for a few days is preferable to one that can operate for only a few hours. The

(18)

-17-devices manufactured to satisfy these needs will not only have to deliver the necessary power, but be small and unobtrusive as well.

Finally, it is promising that new developments in power supply technology have historically been quickly accepted in the marketplace. A prime example of this is the adoption of lithium-ion battery technology. Lithium-ion batteries were introduced 1990, but by 2002 Li-ion batteries already powered an estimated 75-80% of all laptop computers and about 35% of all wireless telephones [2]. This point is further supported upon investigating Figure 1-1, which illustrates the growth in lithium-ion battery sales over the first years of the technology's existence.

30U 250

c 200

0 g 150 100 Cu ) 50 0 1990 1991 1992 1993 1994 1995 1996 1997 Year

Figure 1-1. Sales growth of various rechargeable cell technologies. [5]

1.1.2 Military Applications

When properly outfitted with electronics systems, mechanized military forces are much more effective in outfighting larger but electronically blind adversaries [6]. For this reason, the United States military is pushing for a transition to a digitized battlefield. The electronics systems currently being developed - including communication, navigation, and guidance systems - all require portable, reliable power sources. In addition, the energy source must be

(19)

lightweight, so as not to impede the soldier on the battlefield. Due to their potentially large energy densities, the use of fuel cell systems may be a viable option for satisfying this need.

The performance requirements for military-grade energy devices are much more demanding than those for consumer-grade batteries. For example, batteries currently used by the military must be able to operate in a much larger temperature range, from -400_{C to 60}0_{C rather}

than the 00_{C to} ₅₀0_C_{range required by consumer applications [7]. Furthermore, military}

applications place a much larger premium on size and weight than do consumer applications.

It is expected that lithium-ion battery technology will continue to provide the military with acceptable levels of portable power in the short term [7]. However, significant improvements will be necessary in order to fulfill the military's power needs into the 21st Century. Already, battery/fuel cell hybrid systems have been predicted to play a large role in supplying power to the dismounted soldier in the near future [4, 7].

1.2

Batteries

1.2.1

Battery Basics

A battery is defined as any device that converts chemical energy stored within the battery into electrical energy via an electrochemical oxidation-reduction reaction between an anode and a cathode [8]. While a fuel cell is defined similarly, it is important to note that the reactants in a fuel cell originate from an external location, whereas in a battery, the fuel is an integral part of the device.

(20)

- l

2

F OW(1 o

Flow of c

14-Figure 1-2. Battery schematic.

As depicted in Figure 1-2, a battery is comprised of two electrolytes - the anode (negative) and cathode (positive) - separated by an insulating layer, termed the electrolyte. During discharge, the anode is oxidized and electrons are released. Meanwhile, the cathode is reduced, accepting the electrons from the anode. These electrons can be used to power a load external to the system. To complete the circuit, mobile ions pass through the electrolyte. Depending on the type of system employed, positive ions pass from anode to cathode or negative ions pass from cathode to anode.

1.2.2

State of Battery Technologies

Depending on whether or not the system can be recharged, batteries are labeled either primary (non-rechargeable) or secondary (rechargeable). In order to recharge a secondary cell, current must be passed through the system in the direction opposite that of the discharge current. Primary batteries - ubiquitous on the consumer market - are optimal for portable electronics devices as they are inexpensive, lightweight, and possess a long shelf life. Secondary batteries, while more expensive, are ideal for larger applications as they can provide more power than primary batteries. Batteries are further characterized according to the materials that comprise the

I

power

lectron

supply

El I trC

(21)

anode and cathode. For example, a Zn/MnO2 primary cell is a non-rechargeable battery with a Zn anode and MnO2 cathode.

In characterizing battery performance, two very important measures include specific energy and energy density. Specific energy (W h kg') describes the net energy per unit weight, while energy density (W h m-3_{) indicates the net energy per unit volume.} _{Obviously, it is}

desirable to achieve high specific energy and energy density, thus maximizing the amount of energy that can be derived from a small, lightweight cell. Figure 1-3 illustrates typical values of specific energy and energy density for current battery technologies.

400 Lithium Cylindrica Lithium Coin , 300 .c-Zinc Air >4 Alkaline -

I

200 Silver 0 a) C 100 Me.rirv L Carbon-Zinc 0 Zn-v 100 1000 5000 0 50 100 150 200

Specific Energy (W h kg-1₎ _{Specific Energy (W h kg-}1

)

(a) (b)

Figure 1-3. Modified Ragone plot of energy density vs. specific energy for current technologies in (a) primary

batteries and (b) secondary batteries. [5, 9]

From Figure 1-3a, it can be seen that alkaline batteries - those most popular on the consumer market - represent the median in energy density and specific energy. Lithium batteries have yet to enter the consumer market for a variety of reasons including manufacturing expense and safety. In the case of Zinc Air batteries, a suitable package has yet to be discovered that would allow for its entrance into the consumer market. It should be noted, however, that

1000 500 .-J 5100 w 50

(22)

---zinc air batteries might be more accurately described as fuel cells, since the air used to power the cell originates from outside the device package. In the secondary battery field (Figure 1-3b), lithium batteries again achieve the highest performance. As discussed previously, sales of lithium-ion batteries - the second best performer in the secondary battery field - have steadily increased over the past decade.

1.3

Electric Generators

Rather than deriving electricity from an internally stored fuel, electric generators are used to convert the stored chemical energy of an external fuel into electricity. For example, a heat engine utilizes combustion to extract electrical energy from an externally stored hydrocarbon fuel. Similarly, fuel cells produce energy via reaction (e.g. oxidation) of an external fuel such as hydrogen gas.

Two characteristics suggest that electric generators have the potential to overtake batteries as the leading portable power technology: specific energy and energy density. Typical hydrocarbon fuels used in electric generators can provide energy densities as high as 50 MJ/kg. This is roughly 100 times greater than that of a typical Li-ion battery (-0.5 MJ/kg). Therefore, even at an operating efficiency of only 10% - a typical value for a thermoelectric generator - the generator will provide 10 times more power to the portable device than the battery. Obviously, the advantages offered by the increase in power density are enormous.

(23)

Table 1-1. Energy Properties of Common Fuels [ 101

Fuel Specific Energy Energy Density

(W h/kg) (Wh/L) Hydrogen (gas) 33,300 3 Methanol 5,530 4,370 Ethanol 7,460 5,885 Propane 12,870 6,320 n-Butane 12,700 7,280 Isooctane 12,320 8,504 Ammonia 5,167 3,110

Note: Based on lower heating value

It is important to note that when considering electric generators, one cannot neglect the size or weight of the energy conversion system, as was done in the analysis above. The energy conversion system itself will always contribute to the weight and volume of the total system, thus decreasing the overall specific energy and energy density as illustrated in Figure 1-4.

>. w i11 z w U I-C. -uJ W Generator Weight

Figure 1-4. Plot of electrical energy versus system weight for generators and batteries. Note that the steps in the battery curve represent the incremental increase in system performance with an increase in number of discrete cells. [11]

(24)

From Figure 1-4, it can be seen that in order to minimize the effect of the generator mass on energy density and specific energy, it will be necessary to make the generator as small as possible. Examining the generator curve (blue), one notices that the smaller the generator weight is, the less efficient the generator needs to be in order to meet the same specifications as the battery system, that meeting point represented by the intersection of the red and blue curves. The need to miniaturize the system as much as possible provides further support for the choice to manufacture the system using microfabrication technology. While this choice restricts the possible design geometries, it is justified since it allows for the manufacture of a device on the order of a square centimeter in size.

1.3.1

Fuel Cells

Fuel cell systems have several advantages over other electric generators. Perhaps most importantly, fuel cells have the potential to operate at very high efficiencies (as high as 60%) over a wide range of temperatures. Electrochemical devices such as fuel cells can achieve such high efficiencies because they are not limited by the Carnot efficiency.

Due to the absence of moving parts, there are no frictional losses in a fuel cell system, and the unit operates quietly. Additionally, flame quenching is not an issue in these systems, and heat loss effects are thus reduced. Solid oxide fuel cells can operate on hydrocarbon fuels and do not require hydrogen as the fuel. Not only are hydrocarbons safer to handle, but their use eliminates the need to devise an efficient hydrogen storage scheme. Yet another advantage of fuel cell use is the benign nature of the by-products formed. In many cases, only H20 and CO2

(25)

1.3.2 Microengines

Mechanical engines - such as internal combustion engines and steam turbines - convert

combustible fuels into mechanical or electrical energy via a power cycle. The combustion engine and the gas turbine are two examples of this technology. Mechanical engines are capable of achieving efficiencies of up to 40%, an attractive number for small-scale applications.

There are currently several ongoing research efforts seeking to miniaturize engine technology for portable power generation. Epstein et al at MIT have developed a silicon microfabricated gas tubine engine the size of a quarter [12, 13]. Running on hydrocarbon fuels, these engines on a chip could potentially provide power 10 times as long as today's best batteries. A research team at the University of California Berkeley is developing a stainless steel micromachined rotary engine, with the hopes of producing a device capable of delivering 30 W [14, 15]. Research teams at the Georgia Institute of Technology and MIT have collaborated to produce a microengine/magnetic induction generator coupling that has produced 1.1 W of energy

[16].

While the preliminary results in the field of microengines are promising, their

development poses a difficult technical challenge. The designs are very complex and have yet to

achieve significant yields of successful devices. In the case of the MIT microengine, each of the

individual components work, but they have yet to function simultaneously in a single device [17].

In addition, microengine designs call for the operation of moving parts at high speeds and

temperatures. Maintaining quiet operation and thermal stability of the devices is challenging.

(26)

1.3.3

Thermoelectric and Thermophotovoltaic Generators

Thermoelectric (TE) and thermophotovoltaic (TPV) generators passively generate electricity from the combustion of fuel. In the case of TE generators, a fuel is combusted to locally heat a zone of the reactor. A thermopile bridges the hot and cold zones of the reactor, and electricity is generated via the Peltier-Seebeck effect. In TPV generation, gases are again combusted in a reactor. The resultant radiation is filtered and passed to a low-bandgap TPV cell where it is converted to electricity.

TE generators have been developed by several research teams. Schaevitz et al. at MIT have investigated the use of membrane reactors for TE power generation [18]. The microfabricated device is capable of achieving a thermopile output voltage of up to 7 V, with a thermal efficiency of 0.02%. Sitzki et al. from the University of Southern California have developed a "Swiss roll" burner design capable of generating an electric power output of 0.1 W in a volume of 0.04 cm3 [19].

Thermophotovoltaic generators have been researched since low-bandgap photocell materials became available in the late 1980s. Nielsen et al. at MIT have studied the performance of a TPV generator powered by a suspended-tube combustor [20]. The system is capable of generating up to 1.0 mW of electricity at 0.01% efficiency. The National University of Singapore has developed a micro-TPV system comprised of a silicon carbide emitter, a nine-layer dielectric filter, and a GaSb PV cell array [21]. This device is capable of producing 0.92 W

of power in a volume of 0.113 cm3.

TE and TPV generators allow for the quiet generation of electricity without the involvement of moving parts. While the components can be expensive, the designs are relatively simple. In both cases, however, efficiencies are limited. TE generators are limited by the low

(27)

efficiency of thermal to electric energy conversion, which has yet to exceed efficiencies above 12% at temperatures of 11000_{C [18]. Large-scale TPV systems have achieved efficiencies as}

high as 12.3% [22].

1.4

Fuel Cell Systems

1.4.1

Fuel Cell Basics

A fuel cell is an electrochemical device that continuously converts a fuel into electric energy via reaction with an oxidant. While many types of fuel cells have been developed, this section will focus the three that were identified as options for portable power generation: the polymer electrolyte membrane (PEM) fuel cell, the direct methanol fuel cell (DMFC), and the solid oxide fuel cell (SOFC). Figure 1-5 provides a basic schematic of a hydrogen-powered fuel cell. e" 02,

H20-HO

r S. I. S

I

,•H ₂

0-

H2 I I I porous porous

Cathode Electrolyte Anode

Figure 1-5. Schematic of hydrogen-powered fuel cell operation. From [23].

While the materials of construction vary, the basic design is consistent among fuel cell types: an electrolyte is sandwiched between a porous anode and porous cathode. The fuel - H2

(28)

in the case of Figure 1-5 - passes through the porous anode and reacts catalytically at the anode-electrolyte interface. The electrolyte allows ions to pass through - positive or negative depending on the type of fuel cell - and prevents the passage of electrons which are forced out of the cell to power a load. Meanwhile, the oxidizer - usually 02 from air - travels through the porous cathode and reacts catalytically at the cathode-electrolyte interface. In the case of Figure 1-5, the oxidizer accepts the ions from the electrolyte and the free electrons from the external load, which completes the circuit. The by-product of the process - water and, in the case of direct hydrocarbon processing in solid oxide fuel cells, CO2 - is passed out of the system.

1.4.2

Hydrogen-Fueled Polymer Electrolyte Fuel Cells

The hydrogen-fueled polymer electrolyte membrane (PEM) fuel cell - sometimes called the proton exchange membrane fuel cell - is familiar to most as the fuel cell used to power automobiles [24-27]. The operation of the hydrogen PEM fuel cell is illustrated in Figure 1-5. Hydrogen PEM fuel cells use oxygen from the air as the oxidant. The hydrogen reacts at the anode to form two protons and two electrons. The protons pass through the electrolyte and recombine with oxygen at the cathode to form water.

Anode: H2 -- 2H + 2e- (1-1)

Cathode: 4H+ +02+ 4e- -+ 2H20 (1-2)

Hydrogen PEM fuel cells have a number of advantages. They operate using platinum/carbon electrocatalysts in temperatures ranging from 60-1200C. The low-temperature

operation of hydrogen PEM fuel cells allows for quick start-up times and reduces the risk of injury due to burns. Another advantage of hydrogen PEM fuel cells is that they produce only water as a by-product of their operation. There are also several drawbacks associated with the

(29)

use of hydrogen PEM fuel cells. The polymer membrane - usually Nafion® - must be kept wet

in order to facilitate the efficient transport of protons. Additionally, at temperatures of 1000_C,

platinum is easily poisoned when exposed to ppm levels of CO [28]. For this reason, hydrogen PEM fuel cells are almost always powered by on board ultra-pure hydrogen. The storage of hydrogen poses many difficult challenges as will be outlined later. One alternative is to reform hydrocarbons on board, which would require the purification of the hydrogen production stream.

Hydrogen PEM fuel cells can achieve efficiencies as high as 40-60% [29]. Although hydrogen PEM fuel cells have been mainly developed for mid-scale applications such as automobiles, there has been an acceleration of research focused on the miniaturization of the technology. Lee et al. at Stanford University have developed an integrated series connection of PEM fuel cells using a "flip-flop" connection [30]. The peak power of the system has been reported to exceed 40 mW/cm2. Madou et al. at the University of California, Irvine has developed a micro PEM fuel cell that makes use of pyrolyzed carbon fluidic plates [31 ]. On the commercial side, development of PEM fuel cells has also accelerated in recent years. The Nippon Telegraph and Telephone Corporation has developed a PEM fuel cell small enough to mount directly in a cell phone [32].

1.4.3 Direct Methanol Fuel Cells

Direct methanol fuel cells (DMFCs) are currently the leading fuel cell technology for portable power generation. As the name suggests, DMFCs are powered by a methanol solution. Methanol and water react at the anode to produce C02, six protons, and six electrons. As in

PEM fuel cells, protons are conducted through the electrolyte layer of the DMFC. These protons react with oxygen at the cathode to produce water.

(30)

-29-CH3OH +H20 "- CO2

+ 6H + 6e

Cathode: 12H+ +302 +12e- - 3H20 (1-4)

One advantage of DMFCs is that they are powered by easily stored methanol. Additionally, they require no on-board reforming, thus reducing the complexity of the overall system. Like PEM fuel cells, DMFCs are also operated at low temperatures of 60-1200_{C. There}

are also several limitations associated with DMFCs. DMFCs require water for proper operation, both for the catalytic reaction at the anode and to ensure proper transport of protons through the electrolyte. Water decreases the energy density of the fuel feed. Additionally, the use of water in the electrolyte can lead to "methanol crossover"; as water is transported through the electrolyte, the methanol dissolved in solution accompanies it. Once the methanol reaches the cathode, it can directly oxidize, drastically reducing the cell voltage. In order to prevent methanol crossover, one can either recirculate the water produced at the anode or dilute the methanol fuel to low concentration. Both of these methods reduce the overall efficiency of the cell. In addition to the problems with water, the reaction of methanol at the anode is relatively slow compared to those seen in PEM fuel cells and solid oxide fuel cells.

Despite the challenges outlined above, many large companies including Fujitsu, Hitachi, LG, NEC, Samsung, Sanyo and Toshiba have developed portable electronic devices powered by DMFCs. Smaller companies have also been involved in the development of DMFCs for portable power. MTI Micro (Albany, NY) has developed Mobion" technology that allows DMFCs to run on a 100% methanol feed [33]. Ultracell Power (Livermore, CA) has developed a 40-ounce DMFC capable of delivering 25 W of continuous power [34]. Myriad other companies

(31)

are currently developing DMFC systems, although an in-depth description is beyond the scope of this thesis.

1.4.4

Solid Oxide Fuel Cells

Solid oxide fuel cells (SOFCs) operate at high temperatures, around 500 to 10000C. High temperatures introduce several challenges in the selection of materials. Specifically, it is important to manage the thermal stresses that arise during high temperature operation. The SOFC can be operated using a variety of fuels. Air is supplied to the cathode delivering 02 as the oxidant. 02- acts as the charge carrier in the electrolyte.

The SOFC is an attractive choice for power generation for several reasons. First, the solid oxide electrolyte renders the SOFC more mechanically robust than fuel cells employing liquid electrolytes. Preventing leakage of liquid electrolytes would be particularly difficult in a microdevice. Along the same lines, the SOFC operates using gaseous reactants, which are much easier to handle than the liquids used in direct methanol fuel cells. Most importantly, SOFCs can directly process fuels other than hydrogen - such as hydrocarbons, alcohols, and ammonia - due to high operation temperatures. As will be discussed, the fuel processing step always produces carbon monoxide. Carbon monoxide can be fed directly to an SOFC without poisoning the catalyst, as it does in other fuel cell types, specifically the hydrogen PEM fuel cell. This feature allows for the elimination of a fuel purification step between the fuel processor and the SOFC. Finally, the high temperature operation reduces the sensitivity of the fuel cell system to perturbations from the external environment.

The use of SOFCs also poses design challenges, mostly associated with the high temperature operation. Since the device must operate at temperatures of at least 5000_{C, effective}

(32)

-31-thermal management is critical. Any -31-thermal losses to the environment will directly affect the efficiency of the unit. Furthermore, the high temperature operation raises the issue of materials compatibility and stability.

1.5

Fuel Processing for Hydrogen Generation

1.5.1

Problems with Hydrogen Storage

Ideally, one would prefer to feed hydrogen fuel to the SOFC directly rather than reform a pre-fuel gas to produce hydrogen. Unfortunately, storing hydrogen on-board can be difficult, dangerous, and inefficient. Current research is exploring novel ways in which on-board hydrogen storage can be accomplished. Pressure cylinders have been utilized for large-scale operations, but hardly seem reasonable for fueling portable power sources. Compressing hydrogen into a cryogenic liquid has also been proposed, but the low density of liquid hydrogen (0.07 g cm-3) likely eliminates this option from consideration as well. Research involving active carbon nanotubes had been promising, but recent developments in this field have been scarce. Synthesis of hydrogen from reversible and non-reversible metal hydrides has shown promise recently. Varma et al. at Purdue University have developed a method of producing hydrogen from combustion-assisted hydrolysis of sodium borohydride [35]. This method has been shown to stably generate hydrogen in batch with a yield of 7wt%.

Given the low hydrogen density of existing technologies, storing hydrogen on-board as a fuel source for the SOFC is not an attractive option. This thesis focuses on the development of a method for producing hydrogen on-board from an alternative fuel source. There are many chemicals that can be used as reactants to form hydrogen including hydrocarbons, simple

(33)

alcohols, and ammonia among others. Regardless of the chemicals employed, the process of reacting these fuels to form hydrogen is referred to as fuel processing.

1.5.2 Steam Reforming of Hydrocarbons and Alcohols

The most popular method of hydrogen production currently used in industry is steam reforming of hydrocarbons and alcohols. For hydrocarbons, this reaction proceeds as:

CxH2x+2+ H20 - xCO+(x+2)H2 (1-5)

Reaction 1-5 is highly endothermic and therefore must be carried out at high temperatures. For example, steam reforming of butane is commonly carried out at temperatures of 700 to 10000_{C. To achieve these high temperatures, it is necessary to bum a small portion of}

the inlet hydrocarbon:

CxH2x+2 +

(2(3x

+1)O2 > x CO2 + (x +1)H20 (1-6)

When run in concert, Reactions 1-5 and 1-6 can produce hydrogen autothermally (i.e. without the aid of an external heat source). To maximize hydrogen output, the products of Reaction 1-5 are often further reacted via the water-gas shift:

CO +H20 ++ CO2 + H2 AHo = -41 kJ/mol (1-7)

At the microscale, the use of the water-gas shift is problematic. Normally, it is desired to run this reaction at lower temperatures, thus driving the equilibrium to the right. Operation at low temperature lowers the kinetic rate of the reaction, thus requiring large residence time in the reactor to achieve equilibrium. This large residence time can only be achieved through use of a large reactor, which makes operation at the microscale difficult.

(34)

-33-Similarly, alcohols can be used as fuel for the steam reforming reaction. One advantage to using alcohols is that the reactions are less endothermic than those involving hydrocarbons. Steam reforming of methanol can be carried out at temperatures as low as 2000_{C using}

zinc-oxide supported catalysts. The general reaction for the steam reforming of alcohols is:

CxH

2x+IOH+H 20 -+ x CO2 +(x +

2)H

2 (1-8)

1.5.3

Partial Oxidation of Hydrocarbons and Alcohols

Rather than reacting hydrocarbons and alcohols with steam, an alternative is to partially oxidize the reactants in oxygen gas, as shown in Reactions 1-9a and b:

CxH 2x+2 + 0O2 0 xCO+(x+1)H2 AHo<0 (1-9a)

CxH 2x+IOH +

/02

- x CO2 + (x +l)H 2 AHo <0 (1-9b)

It is important that Reactions 1-9a and 1-9b be carried out using less than a stoichiometric amount of oxygen to avoid forming CO2 in the process. The CO produced in Reaction 1-9a can

either be used as a reactant in the water gas shift to produce more hydrogen, or it can be directly consumed by the SOFC. If another type of fuel cell is used, it is necessary to purify the outlet since the CO will poison the catalyst. One major advantage of the partial oxidation method is that the reaction is exothermic. Therefore, there is no need to generate heat via an additional oxidation reaction in order to drive the reaction. This allows for a much simpler fuel processor design.

(35)

1.5.4

Thermal Decomposition

Thermal decomposition (i.e. cracking) can be carried out using a variety of fuels including hydrocarbons, alcohols (usually methanol), and ammonia. Examples of these three reactions are shown in Reactions 1-10a through 1-10c below:

CxH2x+2

-

C(s) +(x +1)H2

(1-10a)

CH₃OH - CO+ 2H₂ (1-10b)

NH3 -> YN 2 +XH2 (1-10c)

Hydrocarbon cracking (Reaction 1-10a) is an endothermic process, requiring the combustion of 10% of the fuel source to drive the cracking of the other 90%. Without the aid of a catalyst, hydrocarbon cracking requires temperatures in excess of 10000_{C, although the use of}

catalysts can reduce this requirement to as low as 8500_{C [36]. These temperatures are well in}

excess of those required to power the SOFC, so thermal management would become a very difficult task if this reaction were chosen for the fuel processor. Another disadvantage of hydrocarbon cracking is that the reaction produces solid carbon as a product. This carbon will deposit (coke) onto the catalyst, necessitating a catalyst regeneration step between reaction cycles.

Similar to steam reforming and hydrocarbon cracking, methanol cracking (Reaction 1-10b) is also an endothermic process. However, methanol cracking offers several advantages over hydrocarbon cracking. First, methanol cracking does not result in the formation of solid carbon, so frequent catalyst regeneration is not necessary. Furthermore, methanol cracking can be run at temperatures of around 4000_{C with the aid of catalysts [37]. Unfortunately, methanol}

(36)

moles of H2 are produced per mole of methanol via steam reforming while only 2 are produced via cracking. While the water-gas shift reaction could be used with the CO effluent to make up for this effect, the addition is not optimal for work in the microscale.

Finally, ammonia cracking (Reaction 1-10c) is yet another endothermic reaction capable of producing hydrogen gas. This reaction can be run catalytically at temperatures as low as 5000_C. _{Similar to the other endothermic processes, it would be necessary to include an}

exothermic combustion step in order to drive ammonia cracking. This could be accomplished either by combusting part of the ammonia feed or by combusting the unused hydrogen effluent from the fuel cell system (i.e. anode off-gas). Again, this method does not produce solid carbon, and therefore does not lead to problems associated with catalyst coking. However, one would need to deal with the toxicity of NH3. Additionally, ammonia is energetically expensive to

produce.

1.5.5

Current Portable Fuel Processing Technologies

As mentioned previously, the difficulties associated with hydrogen storage have led many research groups to investigate continuous on-board reforming of energy dense fuels to hydrogen. Due to the requirement of at least one high-temperature step in the fuel reforming process, a primary focus of fuel reforming research has been thermal management (which will be address in

Chapter 2).

In recent years, several examples of autothermal portable hydrogen generation have been developed. Leonel Arana fabricated a suspended-tube microreactor which served as a starting point for the research described in this thesis [ 11]. This device - described in more detail later

(37)

[38]. Palo et al have used a stainless steel microreactor to achieve simultaneous methanol combustion and steam reforming of methanol to produce hydrogen [39]. Holliday et al. also used a stainless steel reactor to produce hydrogen from autothermal methanol reforming at efficiencies between 6 and 9% [40]. Horny et al have developed a compact string reactor with catalytic brass wires to autothermally reform methanol [41].

Ganley et al. at the University of Illinois have developed a microreactor employing a ruthenium-impregnated anodic aluminum catalyst to reform 95% of anhydrous ammonia at

6500_C_{to yield 15 seem of hydrogen [42].}

1.6

Thesis Objectives and Approach

The overarching goal of this thesis was to fabricate a microreactor capable of efficiently performing high-temperature fuel-reforming reactions for use in a portable power generator. In order to produce a highly-efficient reactor, one must develop a thorough understanding of the heat loss pathways in the system and the tools that are available to mitigate those losses. Chapter 2 includes a detailed discussion of thermal management strategies in microdevices for portable power generation. The three pathways for heat loss from the system - conduction, convection, and radiation - are described in detail. The effect of designing at small length scales is investigated. Based on this information, general design strategies to minimize heat losses are outlined. Finally, the methods and tools available for fabricating reactors at the microscale are discussed, including the advantages and disadvantages of each.

Using the insight gained from the analysis in Chapter 2, a detailed design of the micro fuel processor was developed. The design consists of a high-temperature reactor suspended from a low-temperature frame by thin-walled glass tubes. The reactor was designed as a single-unit

(38)

-37-capable of being stacked for operation in various configurations. High-temperature operation restricts the materials and mechanical layout of the design. Furthermore, MEMS processing

-the chosen method of fabrication - limits the available geometries. These and other design limitations are discussed in detail in Chapter 3.

The fabrication of the reactor is described in Chapter 4. The microchannels were fabricated using a wet potassium hydroxide (KOH) etch which is inexpensive and compatible with batch processing. KOH etching results in undercutting of convex comers. The strategies used to address this issue are outlined in Section 4.2.1. To join the glass tubes to the silicon reactor, a glass tube brazing procedure was developed. The glass sealant used in this process was stable to temperatures up to 710 oC. The details of the process are discussed in Section 4.2.3. Finally, the fabrication results and post-fabrication packaging techniques are outlined.

Once the reactor was fabricated, a detailed heat transfer analysis was performed on the system. Theoretical steady-state heat losses were estimated for isothermal reactor operation. The results of these calculations are included in Chapter 5.

Once the reactors had been fabricated, combustion tests were performed using several fuels. The results of the reaction testing are described in Chapter 6. Autothermal combustion of hydrogen, propane, and butane were achieved at atmospheric pressure and under vacuum. The temperature profiles of the reactor were measured for each experiment and compared to the heat transfer results from Chapter 5. A variety of reactor failure mechanisms were observed during testing, and they are outlined in Section 6.5.

Finally, a computational fluid dynamics (CFD) model was developed (Chapter 7). The results of the model were compared to experimental data in order to extract kinetic parameters.

(39)

Once it was established that the model was accurate over a variety of flow rates, the results were used to predict the effect of various design alterations. The model yielded insights into the operation of the reactor that could not be obtained directly via testing such as the temperature distribution through the thickness of the reactor.

(40)

-39-Chapter 2 Thermal Management in Devices for

Portable Power Generation

In order for portable power devices to compare favorably to batteries, they must be small, lightweight, and efficient. As mentioned previously, maximum theoretical efficiency is achieved by employing a high-temperature (600-1000 OC) solid oxide fuel cell as the power generator. In order for the system to operate efficiently, the reforming/fuel cell zone must be maintained at high temperatures while isolated from the environment. Given the enhanced heat transfer in microsystems, thermal isolation poses significant challenges.

2.1

Micro Fuel Processor Overview

The purpose of the fuel processor is to convert easily storable fuel to hydrogen. All of the fuels and reforming reactions mentioned in Chapter 1 require at least one high-temperature step. Additionally, endothermic reforming reactions require heat input from an external source, usually from the combustion of a fraction of the feed fuel. Finally, the use of a solid oxide fuel cell requires high temperature operation in order to facilitate the transport of 02_ through the

electrolyte.

In order to compete with batteries, the size and weight of the fuel processor must be small. Miniturization poses two challenges. First, reduction in reactor size limits the residence time of the reactor. Lower residence times generally lead to a reduction in conversion. To compensate, the reactor must be run at higher temperatures when powering endothermic reactions. Additionally, miniaturization leads to enhanced heat transfer. Enhanced heat transfer increases the difficulty of isolating the reactor from the environment since the effectiveness of traditional

(41)

insulating methods is limited at small scales. One advantage of enhanced heat transfer is that it helps maintain temperature uniformity among reactor components.

In order for the reactor to operate at its maximum efficiency, the amount of heat lost to the environment must be minimized. Therefore, when designing the fuel processor, careful consideration must be given to the methods by which the reaction zone is isolated. Figure 2-1 outlines the various mechanisms by which heat can be lost from the system.

Figure 2-1. Schematic of heat-loss pathways associated with a high-temperature fuel processor. Process heat is lost via conduction, natural and forced convection from the exterior of the reaction zone, radiation from the reaction zone, and process heat lost due to forced convection of the process gasses through the reactor.

Figure 2-1 outlines four pathways by which heat can be lost from the high-temperature reaction zone to the ambient environment. Heat will be conducted through the air and through any solid in direct contact with the reactor (e.g. inlet and outlet fluidic connections). The reactor will also lose heat from its exterior faces via natural convection and forced convection (e.g. airflow due to movement of the device). In addition, the external faces of the reactor will emit radiation. Finally, forced convection of heated reactants and products through the channels of

(42)

the reactor will result in enthalpic losses from the reactor exhaust. The design of the fuel processor must incorporate features that minimize the effects of each of these heat loss pathways while preserving the heat exchange between the combustion and reforming units.

2.2

Characteristics of the Microscale

Microscale systems are defined as those with characteristic lengths on the order of 1 to 1000 microns. Microsystems exhibit several unique mechanical and thermal characteristics including: laminar flow over a large range of flow rates, high rates of heat transfer, low thermal inertia, and low mechanical inertia.

The Reynolds number defines the relative importance of intertial and viscous forces in a fluidic system and is written mathematically as follows:

Inertial Forces vp vpL

Re = - _(2-1)

Viscous Forces

p /L

p

In Equation 2-1, v represents the